Mar 29
2009

Disintegrating changes with GIT and Plastic

Posted by Pablo Santos in Version Control

PSantos

Committing code is easy with any version control system even integrating (merging) branches can be unless you're bound to some ancient tool.

Nowadays a very common modern practice with modern SCMs (ok, GIT can do it, Plastic can do it, and I guess that Mercurial can too) is to isolate each change (bugfix, new feature, new experiment, whatever...) on a new branch and then integrate them together on the main (or master for gitters) branch.

So, integration is easy, but what happens when you want to disintegrate a change?

In order to make the post more informative, I'll be explaining the process for both GIT and Plastic SCM.

Suppose I've made some changes on my codebase (I'll start working on a single branch to make things easier) so that it looks like the following:

I've made some changes on the changeset (commit) in the blue square third from the left, and then continued working.

I modified two files on commit 12 (the one in blue) and here are my changes:

And the second file:

And then modified the same files again to make the sample easier to follow on commit 13 as follows:

 

Well, I'd like to get rid of the changes on cset 12, but still keeping the ones at cset 13, which are good changes coming from a different fix.

How should I do that?

Disintegration (how I like to say it), subtractive merge (Clearcase parlance, and also the name of the command in Plastic) or revert (as GIT would say) comes to the rescue!

 

The image shows the current situation, and the commands to create a new commit removing (disintegrating or subtracting) the changes made on cset 12.

In GIT you'll use the revert command while in Plastic you'll use sm, short from subtractive merge.

Subtracting graphically

If you're using Plastic, you can easily subtract a changeset from the branch explorer or the changesets view.

 

It will launch the subtractive merge process, which will show you the elements to disintegrate:

 

If the merge is not automatic (not like the case I described above, which doesn't require any manual operation), the three way merge tool will show up, like in a regular manual merge process.

What's the result?

Well, as expected, just the changes on commit 12 will be removed.

 

Check how the added code in cset 12 is now removed, and the removed one (second file) is there again.

Plastic will create a link between the disintegrated revisions the same way it does after merging, but this time using a special subtractive merge link type which is drawn on red on the 3D version tree as shown below:

 

Working with branches

What happens when we're working with branches?

Consider the following situation where your team have been working on a couple of branches and then merged them back to the main branch.

 

If you take a look at the 3D tree of one of the items involved in the changeset, you'll see the following:

 

Which shows the merge arrows (green) from the two source revisions into the destination ones.

What if now you realize that task001 shouldn't go to the next release? You should get rid of it, using disintegration.

 

A new cset will be created as a result of the operation, and now your trunk is ready to be released, not including task001 changes anymore.

 

Conclusions

Modern SCMs will allow you run potentially complex operations such us subtractive merge, and as you've seen they're really helpful even on a daily basis, to revert mistakes or buggy tasks.

At Codice we focus on making this operations not only doable, but also as simple as possible, avoiding repetitive and error prone tasks, which doesn't mean you don't need to know what happens under the hood.



Comments (2)Add Comment
...
written by Wim Coenen, March 31, 2009
What happens if I later change my mind and redo the merge? Is this a conflict with the subtractive merge?
...
written by Pablo Santos, March 31, 2009
good question! smilies/smiley.gif

Well, in both tools you'll have to "force" the new merge since it will detect it is already done. I mean, it won't be a problem but you'll have to cherry-pick the right cset and redo the changes.

The system, whether it's Plastic or GIT (the first one will do it graphically, as expected) will drive you through the merge process handling manual conflicts if necessary.

Write comment
You must be logged in to a comment. Please register if you do not have an account yet.

busy

Get your FREE Subscription to Dr. Dobb’s Digest today!

Dobbs Code Talk Quick Poll

This time next year, your most important operating system (host and/or target) will be:

Look Who's Code Talking


Eriawan Kusumawardhono
City: Surabaya

Benji Smith
City: Boston

Phuong Le
City: Houston

Joey Burgett
City: Jefferson City

Bill Smith
City: San Diego

Pete Jenney
City: Wilmington

Dobbs Code Talk Tags

.NET abstraction Ada Adobe Agile Ajax algorithm Algorithmic complexity ALM Analogical reasoning Android Anecdotes Apple Application Development AppStore Architecture and Design ARM Artificial Intelligence Artificial Life Assembler Programming Audio files AVX AWK Banking Bazaar Best Practices Blender Books Brain computer interfacing Build C C Programming C Sharp Cartoon Category theory Cellular automata Clojure Cloud Computing Cobol Cocoa Coder Of The Month Cognition as compression Collaboration Common Process/Frameworks Compilers Computational humour Computational narrative Computational politics Computer Science Computers in art computing pioneers concurrency Conferences Consciousness research Contest Contest140 contests CPlusPlus crime CSharp D Programming Data Centers Databases Debugging Delphi Deployment design Design Patterns Digital Signal Processing Distributed Django Documentation DSL dynamic language Eclipse EDA education Emacs Embedded Systems Encryption engineering Erlang Etymology Excel exception handling Facebook Financial computing Five Questions Flash Flash Lite Flex Forth Fortran Fraud FreeBSD Fun Functional Programming gadgets Games Gender Git gnuplot Go Google Graphics GUI hardware Heron High School High-Performance Computing History Holographic reduced representations HTML5 Humanity Humour Hungarian Notation Identity Inkscape Innovation Intel Interview iPhone J2EE Java JavaFX JavaOne JavaScript language engineering Legal lex LINQ Linux Lisp Literate Programming Logic Programming m4 Mainframes Make Mathematica Mercurial Mesh messaging Metaprogramming Microsoft MID Miscellaneous Musings ML Mobile Software Mobility modeling modular programming multicore Music MVC myblog Natural Language Processing Networking Neural networks newspeak Nokia numerical computing Object Rexx ObjectiveC Office Office 2007 Online spreadsheets OOP Open Source Openaccess publishing OpenBSD OpenSolaris Operating Systems Optimization Oracle Pair Programming Parallelism Concurrency Parsing Pascal Patents Patterns Performance Perl PHP Podcast Pop11 Poplog Privacy Processing Productivity Programming Language Implementation Programming Language One Programming language semantics Programming Languages Programming Style Project Management Prolog Psychology Public understanding of science puzzle Python QA Quantum Computing Quotes Rails Realtime recls Requirements Research practice REST Review RIA rich internet applications Robotics Ruby SaaS Software as a service Scala Schadenfreude Science fiction Screencast Scripting SD Best Practices Search Security Semantic Web Silverlight Snobol SOA social Social Networks Society for the Study of Artificial Intelligence a Software Development Methodology and Management Songs and poems Spending Priorities Spreadsheets SQL Startups Statistics Storage String pattern matching Survey Teaching Testing The Business of Programming The Dobbs Challenge The Future Theory Topology Transhumanism Travel on the Job Twitter Types Unix Upgrade Usability Use Cases USENET User Experience User Interface Design Version Control video virtual machines Virtualization Visual Studio Visual Studio Sponsored Post WCF Web Development Windows Windows 7 Windows Live Wireless WOA WPF X Window System yacc

Subscribe to Dr. Dobbs Newsletter

Email:
Dr. Dobb's Update
Delivered twice a week, Dr. Dobb's Update provides unbiased and objective news, commentary and technical features spanning the entire software development marketplace.

Latest Comments

Jonathan's Last Day at Sun
For the 8 years I worked there, it was fantastic. I worked there under McNealy and I have undying admiration for the guy. I only knew Jonathan periphe...
Implementing Thread Local Storage on OS ...
Back in the day, I did a fair amount of work with PThreads. Wonderful design. Some quirks, but basically really, really nice. Although I wrote a lot ...
More Technonecrophilia with Snobol One-L...
Yeah, It's probably identical except for the (embedded) copy number, I would think. Once it became freely distributable, the copy I've been distribut...
More Technonecrophilia with Snobol One-L...
There's a spitbol-3.7-win.exe at http://code.google.com/p/spitbol/downloads/list . I found it via Dave Shield's blog page http://daveshields.wordpress...
Jonathan's Last Day at Sun
Sadness.

The Latest From Our Member Blogs

How To Select Trainees
Written by Joel Wiesen   
01/27/10
Hiring the right trainee can be harder than hiring a trained programmer.  One approach is described at my website: http://www.aprtestingservices.com/business/lpat/
 
Technical Job Interviews
Written by Keith Kerlan   
01/20/10
What is the best way to interview for software developer positions?  I've been on both sides of the job interviewing table, but have been on the interviewee side of some not too  great inter
 
Timers/timeouts in multi-threaded event-loops
Written by Christof Meerwald   
01/03/10
The traditional way to integrate timeout handling (or timers) in (single-threaded) event loops was to just pass the appropriate timeout value to the select/poll/epoll syscall. While this works fine
 
C vs C++
Written by Issam Lahlali   
12/04/09
I think that the debate "C vs C++" will end when the two langages died, and each one have its advantages and inconvenients, the choice of one instead of another depend on the application c
 
Great Jobs at CISCO
Written by Brent Rogers   
11/30/09
Hello! I am a recruiter at CISCO. We have a number of great jobopportunities at CISCO right now. Please take a look at the job links listedbelow and please send me an updated resume if you are interes
 
OK Labs, ST-Ericsson, and the Mobile/Wireless Ecosystem
Written by Steve Subar   
11/17/09
Two weeks ago, OK Labs and ST-Ericsson announced the selection of OK Labs as ST-Ericsson's mobile virtualization partner. To earn this coveted position, OK Labs prevailed in a rigorous evaluation
 
C++ Ninjas Needed in Santa Clara, California
Written by Brent Rogers   
09/30/09
Hello! I am a recruiter at CISCO. Our PostPath teamin Santa Clara is building a new Email SaaS business at CISCO. We are looking forsenior developers with Zimbra expertise to help us accomplish this t
 
Fighting Fragmentation with Mobile Virtualization
Written by Steve Subar   
09/21/09
Last week Motorola and T-Mobile announced the launch of a new and innovative Android-based smartphone, the Cliq. This attractive, feature-rich slider handset happens to build on a chipset and firmware
 
Insights into Router Design: Unit Testing of Networking Protocols
Written by Rajesh Kumar Venkateswaran   
09/07/09
  Unit testing is a software validation methodology through which a programmer tests individual modules or units of source code. If the programmer has been responsible for developing a networ
 
Insights into Router Design: Implementation of Networking Protocols
Written by Rajesh Kumar Venkateswaran   
09/06/09
  Modern data networking consists of a large number of networking protocols, each of which has its own domain of applicability. Some run on end stations (also called hosts), some on enterp
 
Insights and Innovations in Networking
Written by Rajesh Kumar Venkateswaran   
09/05/09
Networking devices such as routers and switches have evolved quite a bit over the past years, both in the service provider network and in the enterprise. It is a challenge to build these devices, bo
 
reddit threads community
Written by Christof Meerwald   
08/30/09
I have just started a threads community over at reddit to cover topics such as multithreading, concurrency and parallel programming. Feel free to join if you are interested. -- cmeerw.org 
 

The Latest From Dr. Dobbs

DDJ